Multi-resolution spectral analysis using Wavelets

Hi, my first posting here. Tried search for any possible topics
already covered that would probaly help me - nothing so far.
Well, I have a sampled audible data that is spoiled by an artifically
produced noise that supposed to have certain spectral pattern. The
noise isn't appearing all the recorded time, but rather for a short
durations (several instances of few tens of msec each along the
recorded data of several seconds), i.e. the recorded data present a
non-stationary signal.
Moreover, the noise pattern itself is suspected to be non-stationary
but rather having most of the spectral aneregy drifting in time
within certain frequency band.

I'm willing to run a multi-resolution spectral analysis to obtain an
estimation of the noise spectral pattern (i.e. its "stamp") in order
of futher robust filtration.
I've quite new to Wavelets, reading the material available online, so
far have quite shallow understanding of the matter, intend to learn
deeply into engineering aspects of the method.
Meanwhile I would be garteful if someone would point me to the link
where a practical multi-resolution spectral analysis is elaborated,
including a brief math background and actual examples (preferably
with MATLAB wavelets toolbox utilization).

Thanks in advance, Alex
.